Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Support multiline examples with DFToVW #2450

Open
jackgerrits opened this issue May 27, 2020 · 2 comments
Open

Support multiline examples with DFToVW #2450

jackgerrits opened this issue May 27, 2020 · 2 comments
Labels
Feature Request New feature requested in system Lang: Python PandasSupport Items relating to VW's interoperability with Pandas

Comments

@jackgerrits
Copy link
Member

There was significant discussion in the original PR about how to support multiline example the the DFToVW system.

In a nutshell:

  • Define a column which specifies an example id, each row that has the same id is part of the same multiline
  • Most multiline formats require different example types - therefore you you must be able to define several formulae which can be selected based on the value of a column.

See the original PR for the full discussion: #2426

@jackgerrits jackgerrits added Lang: Python Feature Request New feature requested in system PandasSupport Items relating to VW's interoperability with Pandas labels May 27, 2020
@etiennekintzler
Copy link
Contributor

Also, since it's a whole new field I think a new class method could be created (as we did for the "from_colnames" convenience method). We then would have :

  • DFtoVW: for the default constructor
  • DFtoVW.from_colnames: for the easy interface
  • DFtoVW.multiline: for the multiline cases

A proper class method for multiline would prevent the default constructor to be clogged with more parameters and would bring better separation of concerns.

@jackgerrits
Copy link
Member Author

Yeah I like the idea of keeping it separate.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Feature Request New feature requested in system Lang: Python PandasSupport Items relating to VW's interoperability with Pandas
Projects
None yet
Development

No branches or pull requests

2 participants